He's right that a very large number of activities will move (have already moved) into the browser. But we also still need desktop software, and we'll need it for a long time.
There may be a day when everything is in the browser, but that's so far off that it's too early to start focusing on that. What will that look like? How will it work? How does it relate to GNOME, as we currently understand it? People are barely beginning to experiment with answering those questions now; it will be a long time before it's a mass phenomenon. And if you start pushing hard on a particular approach this early, you run a greater risk of ending up in a blind alley.
It would be better to focus on where we are now, and get that right. The current GNOME codebase and its derivatives aren't done yet, and there's a lot of room for integrating with the web rather than moving wholesale to it.
And while we wait for the future to get a little clearer, we can spend some time exploring the possibilities. Maybe that exploration is what he was advocating in this talk? I went to the GUADEC site hoping to watch a stream of the talk but it looks like they only have live streams.